Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Grok 4.3

docs.x.ai

Post reply on HN
Print view View original

571–580 of 608 posts

Re: Grok 4.3

#571

Post by bel8 »

gertlabs wrote: post #529

Grok 4.3 is a unique model in our tests. It's one of the fastest models, and its responses are far smaller/token dense than other models with comparable performance. However, its overall coding reasoning ability is not competitive with the big April releases, and neither Grok 4.20 nor Grok 4.3 have been able to significantly push the intelligence frontier since Grok 4. Grok 4.3 is better in agentic workloads, and a f…

Interesting benchmarks. But how is Deepseek V4 Flash significantly better than Pro in the agentic coding benchmarks?

Re: Grok 4.3

#572

Post by sergiotapia »

sroussey wrote: post #541
sergiotapia wrote: post #492

Earlier quoted context omitted.

It's my go to for searches, DIY, personal finance, and more general slice of life AI. Once it is as good as Kimi K2.6 for coding, I will probably use Grok exclusively. It really is the best conversational AI I've used. It has helped me fix a broken fridge, and a broken electrical oven. Literally saved me at least $4k this year. Edit: Also saved me $600 because I did my taxes with it. H&R Block is cooked. Edit 2: Oh s…

Did you do legal filings with it after doing your taxes? Oh my.

what do you mean?

Re: Grok 4.3

#573

Post by ai_fry_ur_brain »

user34283 wrote: post #385
ai_fry_ur_brain wrote: post #142

Earlier quoted context omitted.

Your $200 claude code subscription is a cheap subsidized plan. You're getting like 40k in tokens a year for $2400. A whole lotta people are about to be sad when they realize they bet their competency on that lasting forever.

Luckily inference is cheap, and other providers offer efficient models. It’s only going to get better in the future.

Its not though, just because your favorite CEO or youtuber said it will, doesnt mean it will. Inference is not cheap, you have no idea what you're talking about. Every new chinese model has doubled their prices in the last two weeks

Re: Grok 4.3

#574

Post by gertlabs »

bel8 wrote: post #571
gertlabs wrote: post #529

Grok 4.3 is a unique model in our tests. It's one of the fastest models, and its responses are far smaller/token dense than other models with comparable performance. However, its overall coding reasoning ability is not competitive with the big April releases, and neither Grok 4.20 nor Grok 4.3 have been able to significantly push the intelligence frontier since Grok 4. Grok 4.3 is better in agentic workloads, and a f…

Interesting benchmarks. But how is Deepseek V4 Flash significantly better than Pro in the agentic coding benchmarks?

Pro is smarter in one-shot problems, but it struggles with custom tooling, and spends too much time trying to figure out our harness. We ran a lot of samples, so I can't make excuses for the model. Flash is truly the better option overall, especially considering speed and cost.

Re: Grok 4.3

#575

Post by nojonestownpls »

renegade-otter wrote: post #523
nojonestownpls wrote: post #479

Earlier quoted context omitted.

Do you not use any major provider's AI at all? Because the other big options are from companies actively aiding a genocide (Google), or companies clamouring to be the tools used in future war crimes (OpenAI and Anthropic - the latter only attempted to put weak muzzles on it, they're still heavily involved). Every one of them is involved in actively involved in destroying non-white people's lives and livelihoods, peop…

As I said, I have no illusions about the "morals" of corporations, especially in this post-shame world, but one has to have lines. Musk is a uniquely vile human being who seems to revel in the suffering of others. It's much different from "good business is where you find it".

Yep, large scale murder is just "business is business", but Musk ouchied my feelings with the bad words and that's far worse - that checks out for the current US left attitude.

As a non-white person, I'm far more worried about the danger and damage from openAI and Google, that is real and current. Elon sees us as inferior and isn't quiet about it like most of the rest of the powerful folks are, but "business is business" gets our families killed far more than some tweets do.

Re: Grok 4.3

#577

Post by Barbing »

Barbing wrote: post #175

Grok 4.3 was completed ahead of its CEO’s lesson on this common safety resource: Asked if he knew anything about OpenAI's "safety card," Musk smiled and replied: "Safety card? Why would it be a card?" https://www.axios.com/2026/04/30/musk-openai-safety-grok Low relevancy in spite of cluster size and musical chair gas generators for time being: Later in his testimony, Musk was asked about a claim he made last summer t…

Most controversial comment I’ve ever made that I know of

Re: Grok 4.3

#578

Post by LeSavant »

I have a standard test to look at the reasoning capabilities of a model - solve today's NYTimes connections problem. Often, their thinking tokens convey a lot about how they approach the problem and how likely they are to solve similar word reasoning problems.

Claude 4.7 and Gemini 3.1 Pro have nailed all so far, GPT 5.5 failed miserably. Of the chinese models, Kimi-K-2.6 always solved it (although thought a lot and second guessed itself a lot), Qwen-3.6-Plus often gave wrong answers and GLM-5.1 just spun around endlessly until I had to stop it.

Grok-4.3 also nailed today's puzzle.

Re: Grok 4.3

#579

Post by Petersipoi »

culi wrote: post #526
Petersipoi wrote: post #519

Earlier quoted context omitted.

Have you ever written a comment about how any of the other LLMs are editorializing in favor of the left, and how that's a problem? Because if you have, I'd love to see the evidence of your intellectual consistency. But something tells me you're just doing the same thing that you're calling out

We don't have any proof of LLMs being editorialized in favor of the left. We have clear proof of Grok and we also literally have a White House Executive Order mandating LLMs be editorialized to fight "woke" Your version of reality is exactly skewed to what's actually going on.

[flagged]

Re: Grok 4.3

#580

Post by Petersipoi »

timacles wrote: post #564
Petersipoi wrote: post #519

Earlier quoted context omitted.

Have you ever written a comment about how any of the other LLMs are editorializing in favor of the left, and how that's a problem? Because if you have, I'd love to see the evidence of your intellectual consistency. But something tells me you're just doing the same thing that you're calling out

> about how any of the other LLMs are editorializing in favor of the left I’m sorry come again now. Would you possibly have some examples of this

There have been numerous controversies. Asking ChatGPT if Charlie Kirk / George Floyd are good people, getting completely ass backward answers. Google refusing to generate images of white people, even to the point of making black German Nazis. Absurd biases around asking things related to Trump.

I mean this sincerely. You not knowing any of these examples is a red flag. You need to change your news source.

Post reply on HN